Bringing diversity to recommendation lists – An analysis of the placement of diverse items

Abstract

The core task of a recommender system is to provide users with a ranked list of recommended items. In many cases, the ranking is based one a recommendation score representing the estimated degree to which the users will like them. Up to now research specifically focused on the accuracy of recommender algorithms in predicting the relevance of items for a given user. However, researchers agree that there are other factors than prediction accuracy which can have a significant effect on the overall quality of a recommender system. Therefore, additional and complementary metrics, including diversity, novelty, transparency and serendipity should be used to evaluate the quality of recommender systems. In this paper we will focus on diversity which has been more widely discussed in recent research and is often considered to be a factor which is equally important as accuracy. In particular we address the question of how to place diverse items in a recommendation list and measure the user-perceived level of diversity. Differently placing the diverse items can affect perceived diversity and the level of serendipity. Furthermore, the results of our analysis show that including diverse items in a recommendation list can both increase and sometimes even decrease the perceived diversity and that the effect depends on how the diverse items are arranged.